Professional Documents
Culture Documents
053546
RESEARCH ARTICLE
*Author for correspondence (kclark@wcupa.edu) range of speeds and runners. Theoretically, increased thigh
angular velocity is not only necessary to satisfy the kinematic
K.P.C., 0000-0002-6934-1271
demands of faster running speeds, but may also lead to greater
This is an Open Access article distributed under the terms of the Creative Commons Attribution lower limb vertical velocity at touchdown, which has been
License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use,
distribution and reproduction in any medium provided that the original work is properly attributed. established as a critical factor for generating the large mass-
specific vertical forces required for high speed running (Clark
Received 12 May 2020; Accepted 27 August 2020 et al., 2017; Mann and Murphy, 2018). Potentially, the angular
1
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
velocity of the front swing thigh as it extends during the last Spatial–temporal variables and thigh angular kinematic variables
portion of the flight phase could be directly related to the lower are presented for a single subject (male sprinter) across his individual
limb velocity at touchdown, serving as a contributing factor to range of speeds in Fig. 2, and for the subject population as a whole
vertical force application and running speed. across all subjects and trials in Fig. 3. Spatial–temporal and thigh
Therefore, the aim of this study was to examine thigh angular angular kinematic variables across all subjects and trials (categorized
motion and kinematic factors influencing vertical force application by slow, intermediate and fast speeds) are also presented in Table 1
across a range of speeds. Our first hypothesis was that thigh angular and Table S2.
velocity would demonstrate a positive and linear relationship with For the spatial–temporal variables, contact time (Tc) decreased
running speed across a range of sub-maximal and maximal with increasing speed for individual subjects (Fig. 2A) and for all
intensities. Our second hypothesis was that thigh angular velocity subjects and trials [Fig. 3A; Table 1: F(2,151)=135.1, P<0.001].
would demonstrate a positive and linear correlation with lower limb Flight time (Tf ) was slightly briefer in duration at fast speeds than at
velocity at touchdown across a range of running speeds, lending slow and intermediate speeds for all subjects and trials [Table 1:
insight into the relationship between thigh angular motion and F(2,151)=7.9, P<0.001], although the differences in Tf between
vertical force determinants. These concepts may contribute to the slow and intermediate speeds were not statistically significant
understanding of human running, with practical applications for (Table 1: P>0.8). Step rate (SR) increased across the range of speeds
training interventions aimed at enhancing running performance. for both individual subjects (Fig. 2B) and for all subjects and trials
[Fig. 3B; Table 1: F(2,151)=179.4, P<0.0001]. Step length (SL)
RESULTS increased across speeds for individual subjects (Fig. 2B) and for all
Running mechanics across entire range of speeds subjects and trials [Fig. 3B; Table 1: F(2,151)=67.1, P<0.0001],
The subjects differed greatly in size and athletic background, but although the increases in SL from intermediate to fast speeds were
demonstrated similar modifications to running mechanics across the not statistically significant (Table 1: P=0.17). Across all subjects
range of speeds. Representative data for thigh angular position and trials, the mean distance travelled by the COM during ground
versus time is presented in Fig. 1, with a male recreationally trained contact (Lc) was 0.97±0.01 m. However, Lc increased slightly
athlete at sub-maximal and maximal speeds displayed in Fig. 1A across speeds (Table 1: H=10.4, P<0.01), though the increases in Lc
and C, and a male sprinter at sub-maximal and maximal speeds from intermediate to fast speeds were not statistically significant
displayed in Fig. 1B and D. (Table 1: P>0.9). Analysis regarding vertical velocity of the lower
Biology Open
Fig. 1. Thigh angular position versus time for two representative subjects, with dashed regions of the graph indicating the ground contact phase.
(A,C) Male recreationally trained athlete at sub-maximal and maximal speeds. (B,D) Male sprinter at sub-maximal and maximal speeds. Faster running
speeds were achieved with higher frequencies and greater total amplitudes of thigh angular motion, resulting in greater thigh angular velocities. At top speed,
the slope of the angular position versus time curve during ground contact was steeper for the sprinter (D) than for the recreationally trained athlete (C),
indicating a greater absolute value of the average angular velocity of the stance thigh during ground contact (ωc). Per Eqn 6, greater ωc was a direct
determinant of the faster top running speed attained by the sprinter.
2
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Fig. 2. Kinematic variables for a single representative subject (male sprinter) across his individual range of speeds. (A) Ground contact time, Tc.
(B) Step rate, SR, and step length, SL. (C) Total thigh excursion during ground contact phase, θc, and total thigh excursion from peak extension through peak
flexion, θtotal. (D) Average thigh angular velocity during entire gait cycle, ωavg, and lower limb vertical velocity at instant of touchdown (Ankle Vztd).
limb (ankle marker) at the instant of touchdown (Ankle Vztd) is subjects and trials (Tc=0.62x −0.86 where x is ωavg, R 2=0.90).
presented in the last section of the Results. Per Eqn 6, ωc× leg length (L0) showed a direct positive linear
With respect to thigh angular kinematics, faster speeds were relationship with speed for all subjects and trials with the best fit
generally characterized by higher frequency and greater total line constrained through the origin (ωc×L0=0.94x, R 2=0.91,
amplitude of thigh angular motion (θtotal ), as illustrated by P<0.0001). Also, ωretr increased with speed across all subjects
representative data in Fig. 1. Increases in θtotal occurred with and trials (ωretr=0.80x−2.14, R2=0.71, P<0.0001; Table 1:
speed for individual subjects (Fig. 2C) and for all subjects and trials H=70.2, P<0.0001). Finally, across all subjects and trials, ωavg
(Fig. 3C; Table 1: H=91.6, P<0.0001). Increases in θtotal across the was related to the other two measures of thigh angular velocity, as
range of speeds were due to both increases in thigh flexion (θflex) ωavg had a strong positive relationship to both ωc (ωc=1.35x−0.55
and increases in thigh extension (θext) (Appendix B and Table S2). where x is ωavg, R 2=0.92, P<0.0001) and ωretr (ωretr=0.98x−2.10
Across all subjects and trials, the total excursion angle during where x is ωavg, R2=0.80, P<0.0001).
contact (θc) approached one radian (mean θc=0.97±0.01 radian).
However, there were increases in θc across the range of speeds for Running mechanics across top speeds
individual subjects (Fig. 2C) and for all subjects and trials (Fig. 3C; Kinematic variables were analyzed across all top speed trials and
Table 1: H=71.8, P<0.0001). Increases in θc across the range of with runners grouped by top speed and sex. There were significant
speeds were due to both increases in thigh touchdown (θtd) and in differences in top speed for the fast versus slow subjects when
thigh takeoff (θto) (Appendix B and Table S2). analyzed within sex (Table 2, males: P<0.001, Δ=14.2%; females:
All measures of thigh angular velocity significantly increased P<0.01, Δ=13.2%). Faster top speeds were significantly related to
across speeds. There was a positive and linear relationship between briefer Tc across all top speed trials (Fig. 4A) and when analyzing
absolute average thigh angular velocity during the entire gait cycle fast versus slow runners within sex (Table 2, males: P<0.001,
(ωavg) and speed for individual subjects (Fig. 2D) and for all Δ=21.3%; females: P<0.01, Δ=15.0%). Faster top speeds were also
Biology Open
subjects and trials [Fig. 3D; Table 1: F(2,151)=204.2, P<0.0001]. significantly related to greater SR across the entire subject
Additionally, the average angular velocity of the stance thigh during population (Fig. 4B) and when analyzing fast versus slow runners
ground contact (ωc) increased with speed for all subjects and trials within sex (Table 2, males: P<0.01, Δ=8.6%; females: P=0.04,
[ωc=1.14x−0.89, R 2=0.88, P<0.0001; Table 1: F(2,151)=257.4, Δ=6.5%). Similar relationships were found for SL across all top
P<0.0001]. Per Eqn 4, Tc showed a strong relationship with ωc for speed trials (Fig. 4B) and when analyzing fast versus slow runners
all subjects and trials (Tc=0.002x 2−0.045x+0.36 where x is ωc, within sex (Table 2, males: P=0.002, Δ=5.7%; females: P=0.015,
R 2=0.91), and Tc also had a strong relationship with ωavg for all Δ=6.6%). However, top speed was not significantly related to Tf
3
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Fig. 3. Kinematic variables for all subjects across all speeds (n=154). Best-fit equations and P-values listed here where appropriate, with R2 values
presented in accompanying panels. (A) Ground contact time (Tc=0.78x −0.89). (B) Step rate (SR=0.31x+1.58, P<0.0001) and step length (SL=0.72x 0.49).
(C) Total thigh excursion during ground contact phase (θc=−0.01x 2+0.20x+0.22) and total thigh excursion from peak extension through peak flexion
(θtotal=0.51x 0.56). (D) Average thigh angular velocity during entire gait cycle (ωavg=0.82x−0.08, P<0.0001).
(across all top speed trials: Tf=−0.03x+0.14, R 2=0.07, P=0.10; females: P=0.003, Δ=11.9%). Finally, ωretr was positively and
Table 2, males: P=0.30, Δ=4.3%; females: P=0.60, Δ=2.5%) or Lc, significantly related to faster top speeds across all top speed trials
(across all top speeds Lc=0.01x+0.93, R 2=0.01, P=0.62; Table 2, (ωretr=1.18x−5.36, R 2=0.64, P<0.0001) and when analyzing fast
males: P=0.07, Δ=6.8%; females: P=0.51, Δ=2.2%). versus slow runners within sex (Table 2, males: P=0.014, Δ=28.2%;
For thigh angular kinematics, faster top speeds were females: P=0.011, Δ=32.5%).
characterized by higher frequency and greater θtotal (see Fig. 1C
versus D for representative data). Across all top speed trials, θtotal Lower limb impact velocity, running speed, and thigh angular
was positively and significantly related to faster top speeds kinematics
(Fig. 4C), although the differences in θtotal did not reach Ankle Vztd demonstrated positive and linear relationships with both
significance when analyzing fast versus slow runners within sex running speed and with measures of thigh angular velocity (ωavg and
(Table 2, males: P=0.099, Δ=5.7%; females: P=0.057, Δ=6.2%). ωretr). Ankle Vztd increased across the range of speeds for individual
For all top speed trials, θc was slightly greater than one radian subjects (Fig. 2D) and for all subjects and trials (Fig. 5A; Table 1:
(mean θc=1.07±0.01 radian). However, θc was negatively and H=72.1, P<0.0001). Faster top speeds were positively and
significantly related with speed across all top speed trials significantly related to greater Ankle Vztd across all top speed
(Fig. 4C), and differences in θc were significant for males (but trials (Ankle Vztd=0.29x−0.23, R 2=0.40, P<0.001), although when
not females) when analyzing fast versus slow runners within sex analyzing fast versus slow top speeds within sex, Ankle Vztd
(Table 2, males: P=0.009, Δ=6.6%; females: P=0.057, Δ=3.4%). demonstrated significant differences for females but did not reach
Additional top speed data for θtd, θto, θext and θflex are presented in significance for males (Table 2, males: P=0.118, Δ=13.9%;
Appendix B and Table S3. females: P=0.015, Δ=19.8%). Ankle Vztd was positively and
Biology Open
Across all top speed trials, ωavg was positively and significantly significantly related to ωavg across all subjects and trials (Fig. 5B),
related to faster top speeds (Fig. 4D) and when analyzing fast versus and when analyzed for top speed trials only (Ankle
slow runners within sex (Table 2, males: P=0.002, Δ=14.2%; Vztd=0.37x−0.41 where x is ωavg, R 2=0.52, P<0.001). Finally,
females: P=0.003, Δ=12.8%). Likewise, ωc was positively and Ankle Vztd was significantly related to ωretr across all subjects and
significantly related to faster top speeds across all top speed trials trials (Ankle Vztd=0.30x+0.83 where x is ωretr, R 2=0.75, P<0.0001),
(ωc=0.77x+2.57, R 2=0.58, P<0.0001) and when analyzing fast and when analyzed for top speed trials only (Ankle
versus slow runners within sex (Table 2, males: P=0.003, Δ=13.7%; Vztd=0.23x+1.15 where x is ωretr, R 2=0.54, P<0.001).
4
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
DISCUSSION thigh angular velocity (Figs 1 and 2). This same pattern was
Experimental tests of hypotheses consistent across all subjects and trials, as increases in running
We undertook this investigation to explore the relationship between speed demonstrated a strong positive relationship with thigh angular
thigh angular motion, ground contact mechanics and running speed. velocity (Table 1). This was best represented by the 91% shared
Based on our framework, we hypothesized that thigh angular variance between running speed and ωavg across all trials (Fig. 3D).
velocity would have a strong positive linear relationship with Additionally, across all top speed trials, faster top speeds had a
running speed across the entire range of trials, and across the positive, significant relationship to both ωavg, ωretr and ωc (Table 2),
subjects’ range of top speeds. We also hypothesized that thigh with 74% shared variance between running speed and ωavg across
angular velocity would have a strong positive linear relationship top speed trials (Fig. 4D). When top speed trials were analyzed with
with lower limb vertical velocity at touchdown. We recruited a subjects grouped by speed and sex, all measures of thigh angular
heterogenous group of 40 subjects that included males and females velocity were significantly greater in fast subjects compared to their
across a range of different sizes (height range: 1.52–1.94 m, mass slow counterparts. Comparing fast and slow runners within sex
range: 45.5–117.0 kg), from a variety of athletic backgrounds (Table 2), the percentage differences in ωavg (∼13 to 14%) were
(recreationally trained individuals, intercollegiate team sport nearly identical to the differences in top speed (∼13 to 14%).
athletes, and competitive track and field athletes), and had them Likewise, our data also supported the second hypothesis. Across
run across more than a threefold range of speeds (3.1–10.0 m s−1). a range of speeds from a slow jog to a maximal sprint, Ankle Vztd
Individual subjects ran faster by increasing both the frequency demonstrated a significant positive relationship with speed (Table 1),
and amplitude of thigh angular motion, resulting in an increase in with 68% shared variance between running speed and Ankle Vztd
Table 2. Kinematic variables across top speed trials, with subjects categorized by sex and top speed
Fast males (n=10) Slow males (n=10) Fast females (n=10) Slow females (n=10)
Spatial–temporal variables
Top speed (m s−1) 9.50±0.10† 8.23±0.13 8.06±0.17* 7.06±0.07
Tc (s) 0.105±0.002† 0.130±0.006 0.117±0.004* 0.136±0.004
Tf (s) 0.114±0.004 0.109±0.003 0.123±0.004 0.120±0.004
SR (steps s−1) 4.58±0.06† 4.21±0.09 4.19±0.09* 3.93±0.08
SL (m) 2.07±0.03† 1.96±0.02 1.93±0.03* 1.80±0.03
Lc (m) 1.00±0.01 1.07±0.03 0.94±0.02 0.96±0.03
Ankle Vztd (m s−1) 2.47±0.15 2.15±0.10 2.16±0.12* 1.77±0.08
Thigh angular variables
θc (rad, deg) 1.02±0.02†, 58.6±0.9 1.09±0.02, 62.5±1.0 1.06±0.02, 60.7±1.0 1.10±0.02, 62.8±1.2
θtotal (rad, deg) 1.73±0.04, 98.9±2.3 1.63±0.04, 93.4±2.2 1.65±0.04, 94.3±2.2 1.55±0.03, 88.7±1.7
ωavg (rad s−1, deg s−1) 7.82±0.13†, 448.0±7.7 6.78±0.25, 388.7±14.1 6.88±0.22*, 394.4±12.4 6.05±0.10, 346.9±5.6
ωc (rad s−1, deg s−1) 9.76±0.19†, 559.1±10.8
Biology Open
5
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Fig. 4. Kinematic variables for all subjects across top speed trials (n=40). Best-fit equations and P-values listed here, with R2 values presented in
accompanying panels. (A) Ground contact time (Tc=−0.013x+0.23, P<0.001). (B) Step rate (SR=0.30x+1.78, P<0.0001) and step length (SL=0.10x+1.14,
P<0.0001). (C) Total thigh excursion during ground contact phase (θc=−0.03x+1.28, P=0.014) and total thigh excursion from peak extension through peak
flexion (θtotal=0.07x+1.03, P<0.001). (D) Average thigh angular velocity during entire gait cycle (ωavg=0.77x+0.58, P<0.0001).
(Fig. 5A). Similar results were evident across all top speed trials, as greater than their slow counterparts. Ankle Vztd was also significantly
Ankle Vztd was significantly related to top speed for the entire subject related to thigh angular velocity (ωavg, and ωretr) across the range of
population. When comparing within speed and sex, Ankle Vztd was speeds, with 75% shared variance between Ankle Vztd and ωavg
significantly greater for fast females than slow females (Table 2), and (Fig. 5B) and between Ankle Vztd and ωretr. Similar findings were
although this differential did not reach significance for males, both also evident across top speed trials, as Ankle Vztd was significantly
fast males and fast females had Ankle Vztd that were more than 13% related to both ωavg and ωretr at top speed.
Biology Open
Fig. 5. Lower limb vertical velocity at instant of touchdown (Ankle Vztd) for all subjects and trials (n=154). Best-fit equations and P-values listed here,
with R2 values presented in accompanying panels. (A) Ankle Vztd versus running speed (Ankle Vztd=0.27x−0.02, P<0.0001). (B) Ankle Vztd versus average
thigh angular velocity during entire gait cycle (Ankle Vztd=0.33x+0.01 where x is ωavg, P<0.001).
6
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Fig. 6. Simplified planar representation of the ground contact phase. (A) Geometry of the ground contact leg, including: symmetrical θtd and θto, θc, L0
and Lc. (B) θtd and θto during the ground contact phase. This framework assumes that the leg angle (from ball of foot to hip, θtd or θto in A) is equal to the
thigh angle (θtd or θto in B, respectively).
Framework evaluation 1991; He et al., 1991; Weyand et al., 2010). In this study, θc
While the data strongly supported the hypotheses, it is important to increased slightly across speeds for all subjects and trials (Figs 2C,
note that alternative outcomes were possible. A strong positive 3C, and Table 1), but the mean increase in θc from intermediate to
linear relationship between running speed and ωavg may not have fast speeds was only 0.07 radians (∼4°). Likewise, Lc increased
occurred if the fundamental framework assumptions regarding slightly across speeds (Table 1), although increases in Lc from
ground contact geometry (θc and Lc) were violated. Based on prior intermediate to fast speeds were not statistically significant. Not
evidence it was assumed that θc≈1.0 radian and that Lc≈L0≈1.0 m surprisingly, there was some between-subject variability in Lc, with
(Clark and Weyand, 2014; Farley et al., 1993; Gatesy and Biewener, values differing between the shortest female and the tallest male by
Biology Open
Fig. 7. Simplified representation of thigh angular motion during the entire gait cycle. This representation assumes symmetrical anti-phase flexion and
extension values during ground contact and flight phases. Thigh angular kinematics as a function of time are determined by the parameters of frequency
(f=1/T, where T is the period) and amplitude (A). The figures in inset diagrams a–g illustrate the thigh angular position in correspondence with the angular
motion presented in the graph. The dashed regions of the graph indicate the ground contact phase, with ωc corresponding to the slope of the angular position
versus time curve from touchdown (+0.5 radian) to takeoff (−0.5 radian).
7
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
that the mean θc was 0.97±0.01 radians across all subjects and trials, ωavg, and not just thigh angular velocity in one specific phase of the
and the mean Lc across all subjects and trials was 0.97±0.01 m, the cycle such as flexion or extension. Measuring ωavg provided a
experimental data generally adhered to the assumptions presented in comprehensive representation of the oscillatory frequency and
the framework. Prior research has suggested that humans and other amplitude necessary to achieve a given running speed. Although
bipedal runners are likely constrained to these excursion angles and thigh angular acceleration was not directly reported in this study, the
contact lengths due to leg extensor muscle mechanical advantage measurement of ωavg over the entire gait cycle accounted for the
(Biewener, 1989; Weyand et al., 2010). rapid angular accelerations that occurred as the thigh reversed from
8
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
flexion to extension or vice versa. These vigorous flexion/extension retraction velocities in an open kinetic-chain movement prior to
reversals were more pronounced in faster runners than slower ground contact, combined with a stiff stance limb that allows the
runners at top speed (Fig. 1C,D), and undoubtedly contributed to thigh to extend rapidly in a closed kinetic chain movement
the greater ωavg observed in fast runners than their slow throughout ground contact. We speculate that a relatively stiff
counterparts. lower limb during ground contact is not only imperative for brief
The results presented here also provide insight into the limb contact times and large vertical forces (Arampatzis et al., 1999; Belli
kinematics underlying force application. Numerous prior et al., 2002; Clark et al., 2017; Kuitunen et al., 2002), but also to
investigations have established the relationship between high- allow the thigh to continuously extend at high angular velocities
speed running and mass-specific vertical force (Bundle and throughout ground contact. Any excessive compliance in the lower
Weyand, 2012; Clark and Weyand, 2014; Dorn et al., 2012; limb during contact may hinder rates of thigh extension, prolonging
Kuitunen et al., 2002; McGowan et al., 2012; Weyand et al., 2000, ground contact time and decreasing running speed.
2010), with recent research indicating that greater vertical forces are Our findings align with recent research linking faster running
due in large part to faster vertical velocities of the lower limb at speeds to increased hip joint muscular activation, torque, work and
touchdown combined with a rapid deceleration of the lower limb power (Belli et al., 2002; Bezodis et al., 2008; Copaver et al., 2012;
during initial ground contact (Clark et al., 2014, 2017). In the Dorn et al., 2012; Nagahara et al., 2020; Schache et al., 2011).
present study, the values of Ankle Vztd across a range of speeds However, few studies have examined methods for longitudinal
(Fig. 5A) closely aligned with data from recent publications (Clark improvement of these determinants. Although the effects of lower
et al., 2017; Udofa et al., 2019). Additionally, although correlation body wearable resistance (Macadam et al., 2017) and hip flexor
and causation must be interpreted with caution, our results suggest strengthening (Deane et al., 2005) have been investigated, the best
that increased thigh angular velocities contribute to the greater methods for enhancing thigh angular velocity during sprinting are
Ankle Vztd observed at faster speeds (Fig. 5B). not clearly established, and require expanded investigation.
Collectively, these findings suggest that thigh angular velocity is Conceivably, coaches and athletes could aim to enhance thigh
related to vertical force determinants. This is perhaps best illustrated in flexion angular velocity in open-kinetic chain movements (forward
Fig. 2D, which depicts nearly parallel increases in ωavg and Ankle Vztd swing phase), or thigh extension angular velocity in open-kinetic
across the subject’s range of speeds. Slower speeds are characterized by chain (retraction) or closed-kinetic chain (ground contact)
reduced thigh oscillatory frequencies and amplitudes, resulting in movements. Given the coordinated and reciprocal oscillatory
relatively decreased thigh angular velocities and slower vertical motion generally demonstrated by the thigh segments, any
velocities of the lower limb at touchdown. Faster running speeds longitudinal improvement in thigh extension velocity should
require increased thigh oscillatory frequencies and amplitudes, resulting require corresponding improvement in thigh flexion velocity, and
in greater thigh angular velocities and faster vertical velocities of the vice versa. In other words, increasing only thigh flexion or thigh
lower limb at touchdown. The latter, when combined with a stiff ground extension capability, without concomitant improvement in the other
contact and rapid lower limb deceleration upon ground contact (Clark variable, is likely to limit the runner because the thighs must
et al., 2017), contributes to the greater mass-specific vertical forces complete the powerful scissor-like action in synchrony. Therefore,
required for faster running speeds (Kuitunen et al., 2002; McGowan optimal training interventions likely need to target enhancing thigh
et al., 2012; Weyand et al., 2000). angular velocity in both flexion and extension actions during both
open- and closed-kinetic chain movements.
Future considerations and practical applications
While building on prior findings, this study raises many topics Concluding remarks
worthy of further investigation. First, we intentionally recruited a Here we investigated thigh motion and lower limb vertical velocity
heterogenous group of male and female subjects from a range of at touchdown across a broad range of runners and speeds. As
sizes and athletic backgrounds, and analyzed running mechanics hypothesized, thigh angular velocity had a direct linear relationship
across a threefold range of speeds. Undoubtedly, some of the strong to both running speed and the lower limb kinematics underlying
statistical relationships we found between speed and the kinematic vertical force application. Our results suggest that increases in thigh
variables were due to this heterogenous subject pool and broad angular velocity are not only necessary to match the kinematic
range of speeds. However, we justified our approach based on the demands for high-speed running, but also contribute to the larger
desire to elucidate macro-level determinants of running speed from mass-specific vertical forces necessary to support faster speeds.
slow jogging to maximal sprinting. Although further research is Therefore, interventions aimed at improving running performance
necessary to confirm whether the relationships established here will likely need to elicit an increase in thigh angular velocity through all
generalize to a less diverse group of athletes within more narrow phases of the gait cycle.
ranges of speed (i.e. elite sprinters at top speed), the existing
evidence regarding the thigh angular kinematic determinants of MATERIALS AND METHODS
speed (Mann and Murphy, 2018) suggest that our major findings Framework
may generalize to homogenous subject populations. A simplified planar representation of the ground contact phase is depicted in
Potential applications of these findings for performance Fig. 6. This framework assumes symmetrical touchdown and takeoff angles
improvement in human sprinting are intriguing. Coaching cues during ground contact and that the leg angle (from ball of foot to hip, θtd or
θto in Fig. 6A) is equal to the thigh angle (θtd or θto in Fig. 6B, respectively).
Biology Open
such as ‘whip from the hip’ have been popularized by some well-
The Lc is determined by leg length L0 and θc:
known practitioners, emphasizing a vigorous scissor-like action of
the thighs (Bosch and Klomp, 2005), and these instructions appear uC
Lc ¼ 2 L0 sin : ð1Þ
to have practical merit. Similar to a hammer striking a nail, high- 2
speed running requires fast rotational and tangential velocity prior to For faster running speeds, prior research has suggested that the total
impact combined with a stiff collision upon impact. Likewise, our excursion angle during contact is approximately 1.0 radian for humans and
findings suggest that greater top speeds require fast thigh angular other bipedal runners (Farley et al., 1993; Gatesy and Biewener, 1991;
9
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
He et al., 1991). Thus, for normal contact excursion angles where θc≈1.0 importantly, the equations above dictate that ωc and ωavg must both increase
radian, Eqn 1 demonstrates that contact length is approximately equal to leg for running speed to increase ( per hypothesis 1).
length (Lc≈L0). Furthermore, since horizontal velocity during the flight Enhanced ωavg should be related to an increased lower limb vertical
phase is constant (negating wind resistance), the runner’s forward speed is velocity at touchdown. As the front swinging thigh extends and causes limb
determined by the time it takes the COM to traverse Lc, where Tc is ground retraction during the last portion of the flight phase directly prior to impact
contact time (Weyand et al., 2010): (e.g. Fig. 7A,B,D,E), faster angular velocity of the thigh should result in
faster tangential and vertical velocity of the lower limb in the distal portion
Lc of the leg. Therefore, as ωavg increases in proportion to running speed, this
Speed ¼ : ð2Þ
Tc should also be associated with increases in thigh angular retraction velocity
Because Lc≈L0, and this distance is generally limited to about 1.0 m for (ωretr) prior to touchdown, and both of these measures should be positively
most adult human runners at faster running speeds (Clark and Weyand, related to vertical velocity of the lower limb at touchdown across the range of
2014; Weyand et al., 2010), increases in speed are usually accompanied by running speeds ( per hypothesis 2).
decreases in Tc (Dorn et al., 2012; Nummela et al., 2007; Weyand et al.,
2010). Subjects and participation
Additionally, leg angular velocity is a crucial determinant of locomotor All testing and data collection were completed in one 90 min testing session at
stability and control (Seyfarth et al., 2003). Average angular velocity of the the West Chester University of Pennsylvania laboratory and indoor athletic
stance thigh during ground contact (ωc) is equal to the total contact facility. A total of 40 subjects volunteered and provided written informed
excursion angle divided by ground contact time: consent in accordance with the West Chester University of Pennsylvania
Institutional Review Board. This included 20 males (mean±s.e.m., age: 21.6±
uc
vc ¼ : ð3Þ 0.5 years, height: 1.80±0.01 m, leg length: 0.94±0.01 m, mass: 79.7±3.0 kg)
Tc and 20 females (age: 21.7±0.4 years, height: 1.67±0.02 m, leg length: 0.89±
Eqn 3 can be rearranged so that Tc is expressed as a function of θc and ωc: 0.01 m, mass: 59.0±1.4 kg). Male subjects included nine intercollegiate or
post-collegiate track and field athletes [≤400 m sprints (n=7), horizontal
uc jumps (n=2)], five intercollegiate team sport athletes [soccer (n=2), rugby
Tc ¼ : ð4Þ
vc (n=2), American football (n=1)] and six recreationally trained subjects.
Inserting Eqn 4 into Eqn 2 and substituting Lo for Lc yields Eqn 5: Female subjects included six intercollegiate or post-collegiate track and field
athletes [≤400 m sprints (n=5), 100 m hurdles (n=1)], eight intercollegiate
Lc L0 vc team sport athletes [gymnastics (n=3), soccer (n=2), rugby (n=1), softball
Speed ¼ ¼ ¼ L0 : ð5Þ
Tc ðuc =vc Þ uc (n=1), basketball (n=1)], and six recreationally trained subjects. Complete
subject descriptive characteristics are listed in Table S1. Per inclusionary
Assuming a total excursion angle θc equal to 1.0 radian, running speed is
criteria, all subjects were healthy and regularly active (defined as exercising
directly related to L0 and ωc:
three or more times per week) at the time of testing.
Speed
Speed ¼ L0 vc or ¼ vc : ð6Þ Testing procedures
L0
After subjects reviewed and signed consent forms, experimental procedures
Furthermore, harmonic oscillatory thigh motion has been observed were as follows. Subjects were provided with standardized compression
during high-speed running in humans (Novacheck, 1998; Sides, 2015), clothes and running track flats (Nike Waffle Racer version nine, Beaverton,
with contralateral limbs exhibiting reciprocal anti-phase flexion and OR, USA). Subjects were measured for height using a standard measuring
extension during bipedal gait (Blum et al., 2014). Fig. 7 depicts a tape and weighed on a digital scale (Supac Model EB-8008, Shanghai,
simplified example of thigh angular motion that assumes symmetrical China). Subjects then performed a full-body warm up including jogging,
anti-phase flexion and extension values during ground contact and flight skipping, dynamic stretches and sub-maximal sprints. Next, for the purposes
phases. Because this framework is based on equations of harmonic of motion capture, subjects wore reflective markers placed on the heel and
motion, thigh angular kinematics as a function of time are determined by ball of the foot on the lateral aspect of the running shoe, as well as on the
the parameters of frequency ( f=1/T, where T is the time period) lateral aspect of the ankle, knee, hip and shoulder (lateral malleolus,
and amplitude (A). Thigh angular position as a function of time is lateral femoral condyle, greater trochanter and acromial process, respectively).
displayed in Eqn 7: There were 12 reflective markers total, six on each side of the body. A still
frame motion capture recording of each subject standing in the center of the
uðtÞ ¼ A sinð2p f tÞ: ð7Þ field of view was completed prior to the testing to serve as a reference for
Thigh angular velocity as a function of time is displayed in Eqn 8: kinematic analyses. Measurements of leg length from greater trochanter to
ground were determined from the still frame motion capture recording, with an
du average leg-length value determined using measurements of right and left leg.
vðtÞ ¼ ¼ A ð2p f Þ cosð2p f tÞ: ð8Þ
dt For the experimental testing, subjects completed 40 m running trials, and
Graphically, the ground contact phase is represented by the dashed lines data were captured and analyzed on four trials over a range of speeds.
in Fig. 7, from touchdown (+0.5 radian) to takeoff (-0.5 radian). Thus, ωc Although the subjects were not directly paced by the investigators while
is equal to the slope of the angular position versus time curve during the running, they were instructed to complete the trials at progressively faster
ground contact phases (slope of the dashed line in Fig. 7B,C,E and F). speeds, with the last trial being maximal. All trials were completed in a
The steeper the slope of the angular position versus time curve in running lane in an indoor athletic facility with a multi-purpose floor. The
Fig. 7B,C,E,F, the greater the absolute value of ωc, and the faster the running lane was 60 m in total length, allowing the subjects 20 m to safely
running speed (normalizing for L0, per Eqn 6). decelerate and stop after completing each 40 m running trial. Subjects
The figures and equations above indicate that thigh angular velocity must began each trial in an upright, ‘two-point’ stance with the preferred leg
Biology Open
be regulated to maintain a constant forward running speed. It has been forward and started at their own initiative. For the trials completed at less
suggested that the limbs function as harmonic oscillators (McGeer, 1990), than maximal intensity, subjects were instructed to gradually accelerate to
which implies that ωc is related to the average of the absolute value of thigh a cone placed at 25 m, and then to run at a constant speed from 25 m
angular velocity during ωavg. Because of the continuous harmonic motion of through 40 m. For the maximal effort trials (interchangeably termed ‘top
the thighs, it would be expected that a faster ωc would correspond to a speed’ trials), subjects were instructed to perform an all-out acceleration
proportionately faster ωavg. From a mathematical standpoint, Eqns 7 and 8 from the beginning of the sprint, and continue at full speed all the way
imply that ωc and ωavg can be improved by increasing either f or A, or through the finish line at the 40 m mark. Subjects were allowed complete
increasing some combination of both (see examples in Fig. 8). Perhaps more recovery between trials.
10
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Data collection and analysis Additionally, several thigh angular kinematic variables were determined,
Three-dimensional kinematic data were recorded using an eight-camera including: thigh angular position at touchdown (θtd), thigh angular position
motion capture system collecting at 200 Hz (OptiTrack Prime 13 cameras at takeoff (θto), total thigh excursion during the ground contact phase (θc, the
with Motive software from NaturalPoint, Corvallis, OR, USA). The field of sum of θtd and θto), peak thigh extension during flight (θext), peak thigh
view captured by the motion capture system was eight meters in length, from flexion during flight (θflex), total thigh excursion from peak extension
31–39 m in the running lane. This field of view was selected to ensure that through peak flexion (θtotal, the sum of θext and θflex), average thigh angular
data could be collected for one complete gait cycle for even the tallest, fastest retraction velocity measured from peak thigh flexion until touchdown ωretr,
subjects with the longest stride lengths. Prior to data collection, the capture average thigh angular velocity during the ground contact phase ωc, and
volume was dynamically calibrated using a wand with reflective markers at average thigh angular velocity during the entire gait cycle ωavg. Thigh
known spacings. After data collection, motion capture files were exported angular kinematics were quantified in an absolute frame of reference, with
and uploaded to Microsoft Excel, where the data were up-sampled to the thigh angle compared to vertical, and not relative to trunk. Thigh angular
1000 Hz using linear interpolation and post-filtered using a low-pass, velocities (ωretr, ωc, and ωavg) were determined by calculating the derivative
fourth-order, zero-phase-shift Butterworth filter with a cutoff frequency of of the thigh segment angular position versus time data, with angular
25 Hz (Winter, 1990). Positional data from the 12 markers were used to form velocities measured as absolute values. As with the spatial–temporal
a seven-segment model, including foot, shank and thigh on both legs, and a variables, reported thigh angular kinematic values were trial averages from
head-arms-trunk segment (Winter, 1990). one complete gait cycle, with trial averages determined from both right and
In addition to measuring running speed using motion capture data left limbs.
(described below), running speed was verified with a radar gun (Stalker ATS
II; Applied Concepts Inc., Plano, TX, USA). The radar data were collected at Statistical analysis
47 Hz and exported to Microsoft Excel for analysis. Speed versus time data To examine the relationship between running speed and the kinematic
were fit to a mono-exponential equation using an iterative least-squares variables of interest, the data were analyzed across all subjects and trials. Six
regression routine (Chelly and Denis, 2001; Samozino et al., 2016). Speed of the 40 subjects had sub-maximal trials with motion capture data that were
versus time data were integrated to determine distance versus time, and then not usable due to marker occlusions. These trials were discarded, resulting in
used to calculate average speed from 31–39 m. Average running speed from n=154 total trials included in the data analysis. Across all subjects and trials,
31–39 m, as determined by the radar data, showed a high level of agreement the relationship between running speed and each of the kinematic variables
compared to the motion capture method (R 2=0.99 and mean absolute error was evaluated using either simple linear regression or nonlinear regression
<0.15 m s−1, see Appendix A and Fig. S1). ( power or second-order polynomial equations) to generate a best-fit
With regard to collecting top speed data in the field of view from 31– equation and coefficient of determination (R2), with x representing
39 m, prior research has illustrated that team-sport athletes approach running speed unless otherwise noted. Furthermore, differences in each of
peak speed by 30 m into a sprint (Clark et al., 2019; Mendiguchia et al., the kinematic variables were analyzed across speeds with trials divided into
2016). Although elite sprinters may not attain peak speed until 60 or three categories based on percentage top speed, calculated for each subject
70 m into a 100 m dash (Krzysztof and Mero, 2013), split-time data from as a percentage of his or her fastest trial. These categories were slow (<75%
100 m competitions has indicated that even the world’s fastest sprinters top speed, n=44 trials), intermediate (75 to 93%, n=48 trials), and fast
have reached greater than 94% top speed by 30–40 m (Krzysztof and (>93%, n=62 trials). The normality of data was determined using the
Mero, 2013). Therefore, we deemed the 31–39 m field of view D’Augustino and Pearson omnibus normality test or the Shapiro–Wilk test.
appropriate to capture top speed for the heterogenous group of Parametric data were analyzed using 3x1 analysis of variance (ANOVA) and
subjects participating in this study. To assess the possibility that faster Tukey’s post-hoc tests, and non-parametric data were analyzed using the
subjects in this study were still accelerating through the motion capture Kruskal–Wallis test (H statistic) and Dunn’s multiple comparisons tests.
field of view during top speed trials, speed versus distance data from the In addition to the aforementioned analysis, top speed trials (n=40, one
radar were examined to ensure that subjects ran at constant speed in the trial per subject) were independently analyzed as a separate sub-set of data,
motion capture zone. Constant speed was operationally defined as with simple linear regression used to determine the relationship between
changes in speed ≤0.3 m s−1 from 31–39 m, and all trials satisfied this each kinematic variable and top speed. To provide further insight into these
criterion. top speed trials, runners were grouped by sex (male and female) and top
speed (the faster ten subjects versus the slower ten subjects for each sex). For
Measurements each kinematic variable, separate independent t-tests were used to analyze
Based on the kinematic data, spatial–temporal and thigh angular kinematic fast versus slow males and fast versus slow females. This analysis was
variables were quantified for all trials. Spatial-temporal variables included: selected instead of a 2×2 ANOVA (sex by speed) because the primary
running speed, Tc, Tf, SR, SL, Lc and Ankle Vztd. All reported values are comparisons of interest were kinematic differences between fast and slow
trial averages from both right and left limbs during one complete gait cycle. runners analyzed within sex. For each data set, the normality of data was
Instantaneous calculations of COM position and velocity were quantified determined using the D’Augustino and Pearson omnibus normality test
using the positions of the 12 reflective markers and segment information or the Shapiro–Wilk test. Parametric data were analyzed using unpaired
from Winter (1990). Running speed was quantified from average COM two-tailed t-tests, and non-parametric data were analyzed using
horizontal velocity in the 31–39 m field of view. With regards to Mann–Whitney tests. For the top speed trials, absolute percentage
determining contact and flight time (Tc and Tf ), all trials were reviewed difference was also used to express the magnitude of difference between
using the motion capture software (NaturalPoint, Corvallis, OR, USA), with fast versus slow group means for each variable, calculated as:
instances of touchdown and takeoff determined from visual inspection of the
heel and ball marker vertical position relative to the running surface. This jmeanone meantwo j
method was validated compared to an in-ground laboratory force plate using absolute percentage difference ¼ 100:
fðmeanone þ meantwo Þ=2g
a separate sub-sample of subjects, and demonstrated a high level of
agreement with the force plate measurements (R 2=0.99 and Tc mean All data are expressed as mean±standard error of the mean (s.e.m.). The a
priori threshold for all significance tests was set at α=0.05. Statistical
Biology Open
absolute error <0.004 s, see Appendix A and Fig. S2). SR was calculated as
the inverse of the sum of ground contact time and flight time, or SR=1/ analyses of thigh angular kinematics were calculated in units of radians,
(Tc+Tf ). SL was calculated from running speed and step rate, or SL=Speed/ although for enhanced clarity, units of degrees are also presented where
SR. Per Eqn 2 and Weyand et al. (2010), Lc was calculated by multiplying relevant in Figs 2–7 and in Tables 1, 2, S2, S3. Power analyses for regression
running speed and ground contact time, or Lc=Speed×Tc. The absolute and ANOVA were completed using G*Power (version 3.1.9, Kiel,
value of vertical velocity of the ankle marker at the instant of touchdown Germany), based on α=0.05, β=0.8 and moderate effect size. All other
(Ankle Vztd) was used as a proxy for lower limb impact velocity ( per Clark statistics were completed using Microsoft Excel and GraphPad Prism
et al., 2017). software (version 8, San Diego, CA, USA).
11
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
Appendix A Kinematic data were analyzed to determine the spatial and temporal variables
Validation of running speed associated with touchdown and takeoff. All trials were reviewed using the
Measurement of running speed using the motion capture method was motion capture software (NaturalPoint, Corvallis, OR, USA), with instances of
verified with a radar gun (Stalker ATS II; Applied Concepts Inc., Plano, touchdown and takeoff (and thus total Tc) determined from visual inspection of
TX, USA). During all trials (n=154), the radar gun was placed on a tripod the heel and ball marker vertical position relative to the running surface. This
at a height of 1.0 m, at a distance behind the starting line of 5.0 m. Radar visual inspection method was validated compared to the force plate, which was
data were collected at 47 Hz, and then exported to Microsoft Excel for used as the standard reference for measuring ground contact time (defined as the
analysis. Speed versus time data were fit with a mono-exponential equation time when vertical force measured by the force plate exceeded 20 N). Compared
using an iterative least-squares regression routine (similar to Chelly and to force plate data, this kinematic visual inspection method determined
Denis, 2001; Samozino et al., 2016). This equation models a runner’s touchdown with a mean absolute error of 2.1±0.2 ms, takeoff with a mean
speed versus time curve as follows: absolute error of 2.7±0.2 ms, and Tc with a mean absolute error of 3.4±
0.3 ms (mean±s.e.m.). Contact times measured by the force plate are
speedðtÞ ¼ speedmax ð1 eðt=tÞ Þ, ðA1Þ
plotted against those determined by the motion capture visual inspection
where speedmax is the runner’s maximal speed and τ is the time constant. method in Fig. S2.
Speed versus time data was then integrated to calculate distance versus
time: Appendix B
ð Additional thigh angular kinematic results from all subjects and trials
distanceðtÞ ¼ speedðtÞdt, ðA2Þ Increases in θtotal across the range of speeds were due to both increases in
θflex (θflex=0.102x+0.273, R 2=0.51, P<0.0001; Table S2: H=45.2,
where distance is the horizontal position of the runner from the starting P<0.0001) and increases in θext [θext=0.020x+0.367, R 2=0.06, P=0.003;
line as a function of time. After distance versus time was determined, Table S2: F(2,151)=16.0, P<0.0001], though increases in θext from
average speed from 31–39 m was calculated. Average speed from 31– intermediate to fast speeds were not significant (Table S2: P=0.18).
39 m, as determined by radar data, showed a high level of agreement Increases in θc across the range of speeds were due to both increases in θtd
compared to the motion capture method, with an absolute error of 0.11 [θtd=0.033x+0.335, R 2=0.33, P<0.0001; Table S2: F(2,151)=29.5,
±0.01 m s−1 (mean±s.e.m.). Average speed measured by the radar gun is P<0.0001] and increases in θto (θto=0.30x 0.16, R 2=0.03; Table S2:
plotted against the motion capture method in Fig. S1. H=13.63, P<0.01), though increases in θto from intermediate to fast
speeds were not significant (Table S2: P>0.9).
Validation of ground contact time
To complete data analysis, it was necessary to develop an accurate method Additional thigh angular kinematic results from top speed trials
for determining the instances of touchdown, takeoff, and total ground Across all top speed trials, θflex was positively and significantly related to
contact time (Tc) using motion capture data. To accomplish this, a sub- faster top speeds (θflex=0.132x−0.023, R2=0.38, P<0.0001), although
sample of subjects completed running trials over an in-ground laboratory differences in θflex did not reach significance when analyzing fast versus
force plate with synchronized force and motion capture data. This sub- slow runners within sex (Table S3, males: P=0.062, Δ=16.5%; females:
sample consisted of ten subjects (mean±s.e.m.: seven males, age: 24.9± P=0.059, Δ=13.0%). Across all top speed trials, θext was negatively and
2.3 years, height: 1.77±0.03 m, mass: 89.5±4.9 kg; three females, age: significantly related to faster top speeds (θext=−0.057x+1.044, R 2=0.21,
21.0±0.6 years, height: 1.64±0.01 m, mass: 55.2±3.6 kg) who volunteered P=0.003), although differences in θext did not reach significance when
and provided written informed consent in accordance with the local analyzing fast versus slow runners within sex (Table S3, males: P=0.117,
university Institutional Review Board. Δ=16.7%; females: P=0.51, Δ=5.1%). Across all top speed trials, θtd was not
During a single 60 min testing session, subjects completed 12 trials in a significantly related to top speed (θtd=0.018x+0.459, R2=0.05, P=0.159) or
35 m laboratory runway with the force plate embedded at the 20 m point when analyzing fast versus slow runners within sex (Table S3, males: P>0.9,
of the runway. All procedures related to subject clothing/footwear, Δ=0.0%; females: P>0.9, Δ=0.1%). Across all top speed trials, θto
measurements, warm-up, and motion capture marker placement were was negatively and significantly related to faster top speeds
identical to those described in the Materials and Methods. Although subjects (θto=−0.043x+0.817, R 2=0.21, P=0.003), although differences in θto did
were not directly paced by the investigators while running, they were not reach significance when analyzing fast versus slow runners within sex
instructed to complete the trials at progressively faster speeds, with the last (Table S3, males: P=0.09, Δ=15.7%; females: P=0.37, Δ=7.4%).
three trials being maximal. Subjects lined up in an upright stance 20 m
behind the force plate and, at their own initiative, ran forward and stepped on Acknowledgements
the force plate during the course of a running stride. Subjects were instructed The authors thank Tyler Whitacre, Sabrina Mangeri, Samantha Jenks, Sean
to run naturally as they ran over the force plate. If the subject missed or Leonard, Ronald Mungin, Andrew Slater, Nick Jachwak, Kyle McCormick, Dylan
partially struck the force plate, or visibly altered their mechanics to strike the Ferron, and Marshall Lane. The authors also thank the subjects for their
plate, the data obtained from the trial were discarded. A total of 120 total participation, especially the WCUTF program and Jason Kilgore.
trials were completed, yielding 95 valid trials with synchronized force and
motion capture data where the subject fully struck the force plate without Competing interests
altering normal running mechanics. The authors declare no competing or financial interests.
Data were synchronously collected with an in-ground force plate (Kistler
5691A, 0.6 m×0.9 m, Winterthur, Switzerland) and an eight-camera motion Author contributions
capture system (OptiTrack Prime 13, Corvallis, OR, USA). Cameras were Conceptualization: K.P.C.; Methodology: K.P.C.; Validation: K.P.C., C.R.M.; Formal
placed around the force plate and provided a 6 m field of view centered on the analysis: K.P.C., C.R.M., D.J.S.; Investigation: K.P.C., C.R.M., D.J.S.; Data curation:
force plate. Prior to data collection, the capture volume was dynamically K.P.C., C.R.M., D.J.S.; Writing - original draft: K.P.C., C.R.M., D.J.S.; Writing - review
& editing: K.P.C., C.R.M., D.J.S.; Supervision: K.P.C., D.J.S.; Project administration:
calibrated using a wand with reflective markers at known spacings. The force
K.P.C., D.J.S.
Biology Open
plate data were collected at 1000 Hz and the motion capture data at 200 Hz.
Force plate data were post-filtered in Microsoft Excel using a low-pass, fourth-
Funding
order, zero-phase-shift Butterworth filter with a cutoff frequency of 25 Hz This research received no specific grant from any funding agency in the public,
(Winter, 1990). All motion capture files were analyzed in an identical manner commercial or not-for-profit sectors.
to that presented in the Materials and Methods. Additionally, trials were
recorded using a high-speed video camera (Apple iPad 9.7, Apple USA, Supplementary information
Cupertino, CA, USA) to ensure that the foot fully contacted the force plate and Supplementary information available online at
that no part of the foot landed off the force plate. https://bio.biologists.org/lookup/doi/10.1242/bio.053546.supplemental
12
RESEARCH ARTICLE Biology Open (2020) 9, bio053546. doi:10.1242/bio.053546
in sprint running. Med. Sci. Sports Exerc. 34, 166-173. doi:10.1097/00005768- movements J. Appl. Physiol. 89, 1991-1999. doi:10.1152/jappl.2000.89.5.1991
200201000-00025 Weyand, P. G., Sandell, R. F., Prime, D. N. L. and Bundle, M. W. (2010). The
Macadam, P., Cronin, J. B. and Simperingham, K. D. (2017). The effects of biological limits to running speed are imposed from the ground up. J. Appl.
wearable resistance training on metabolic, kinematic and kinetic variables during Physiol. 108, 950-961. doi:10.1152/japplphysiol.00947.2009
walking, running, sprint running and jumping: a systematic review. Sports Med. 47, Winter, D. A. (1990). Biomechanics and Motor Control of Human Movement, 2nd
887-906. doi:10.1007/s40279-016-0622-x edn. New York: John Wiley & Sons, Inc..
13